Bone and lung are the two most frequent sites of metastatic recurrence in breast cancer. In work supported by this grant we selected organ-specific metastasis cell subpopulations from the breast carcinoma cell line MDA-MB-231, and derived a bone metastasis gene expression signature (BoMS) and a lung metastasis signature (LMS). We functionally validated many of these genes as mediators of bone or lung metastasis. With W. Gerald and A. Olshen, we showed the ability of these signatures to predict site of relapse in cohorts of primary breast tumors. We are defining the functions that are fulfilled by these genes, including roles of BoMS genes IL11 and OPN in osteoclast activation during osteolytic metastasis, and roles of LMS genes Epiregulin, COX2, MMP1 and MMP2 in tumor neovascularization and metastatic extravasation. With R. Benezra and H. Varmus we are defining the role of LMS gene ID1 in conferring tumor-initiating capacity to metastatic breast cancers of the basal subtype. Based on these insights, we have proposed with L. Norton the "tumor self-seeding" hypothesis of how certain metastatic functions may underlie the association of large tumor size, rapid growth rate, and metastatic behavior often seen in clinical cancer. Additionally, we have defined mechanisms that allow breast cancer cells to avert the tumor-suppressive action of TGFR. Using mouse models developed under P01-CA94060, we demonstrated roles of the TGFIJ-Smad pathway in breast cancer metastasis to lung and bone. These models and target genes provide relevant assay systems to test the effectiveness of existing and novel drugs against breast cancer metastasis. Building on this progress, in the next grant period we are planning to identify new bone and lung metastasis genes using primary malignant cells from fresh pleural effusions (Aim 1). We will seek to define functional and therapeutically relevant interactions between validated metastasis genes (Aim 2). We will identify clinically relevant genes mediating the pro-metastatic action of TGFB in lung and bone (Aim 3). Furthermore, we will experimentally test the "tumor self-seeding" hypothesis (Aim 4). Built into each of these specific aims is the development and application of relevant assay systems to test the effectiveness of therapeutic agents under study in the Program Project. In sum, we are proposing to continue on the pace of discovery in breast cancer metastasis that was initiated during the previous grant period by this cooperative research effort.